We will present recent improvements to our approach for diversity-based combinatorial library design (J. Med. Chem. 1995, 38, 1431). These include new templates and whole molecule descriptors, ``binned'' designs which sample groups from different predefined structural or property classes, and a faster (~ 2 orders of magnitude) multidimensional scaling algorithm. We will also present an approach for structure-based combinatorial library design and compare its practical use in lead discovery vs. lead optimization problems.

Joint work with Eric Martin, David Spellmeyer and Roger Critchlow.

Back to Workshop Schedule

1996-1997 Mathematics in High Performance Computing